Ludvig Renbo Olsen and Martin Nielsen of UNSILO will be participating in the forthcoming third Springer Nature Hack Day (Berlin, 26-27 April). We spoke with Martin Nielsen:
Are you looking forward to the event, Martin?
Yes, it’s a chance to try out a few ideas in a collective way. We are going to study concept embeddings in SpringerLink Biomedicine articles to see if word order makes a difference. You may be familiar with the “bag of words” approach to examining documents, which just takes all the words in the document as if they were gathered and put into a big bag, without taking word order into account. For this event, we are going to try preserving the original order of words in a document using a recurrent neural network, to see if the word order improves the results. At the event, we’ll look at the words in sequence to see what results we get.
What are the benefits?
At the Hack Day, everyone benefits – we all discuss what we are going to do, and at the end of the day everyone presents what they have done. You get a chance to meet Springer people and others collaborating with them.
Isn’t it rather challenging having to build and complete something in a single day, in front of your peers?
I’ve done rather similar projects while at university, but of course there it was being looked at from a theoretical viewpoint. Here we are seeing if it makes a practical difference to the results. Ludvig and I have already agreed the approach we will try, and we will repurpose and modify some of our existing code. In any case, I enjoy the pressure!